Structural and Functional Investigation of Two Splicing Factors: Human hnRNP G and yeast Npl3

Abstract

The genetic information stored in the DNA is transferred by a first step of transcription to RNA. The transcribed pre-messenger RNA (pre-mRNA) is subjected to a series of maturation steps before it is translated into functional protein. One of the major maturation steps is constitutive splicing, which consists in the removal of non-coding sequences (introns) within the RNA and the ligation of coding sequences (exons). Interestingly, the pre-mRNA can be spliced differently by the inclusion or exclusion of specific exonic or intronic elements resulting in different mRNAs from a single gene. This process known as alternative splicing allows the modulation of the message conveyed by each gene. The resulting diversity of messages provides eukaryotic cells some flexibility to react to variations in intra and extracellular conditions. Splicing is subjected to very tight regulation steps to ensure that the correct pieces of spliced RNA are ligated together and give a meaningful mRNA. This control is ensured by two major classes of proteins, namely serine/arginine-rich (SR) proteins and heterogeneous nuclear ribonucleoproteins (hnRNPs). The splicing outcome will depend on their competition to bind the pre-mRNA at a precise position that will allow them to either stimulate or inhibit the splicing reaction. This action is achieved by influencing the recruitment of the nuclear machinery responsible for the splicing catalysis, the spliceosome. In the Spinal Muscular Atrophy (SMA) disease, patients lack a functional copy of Survival of Motor Neurons 1 (SMN1) gene and retain only a closely related gene SMN2 that does not allow the splicing of exon 7 leading to the production of truncated, and non-functional SMN proteins. One of the main activators of SMN2 exon 7 splicing is hnRNP G. We solved the NMR structure of the RNA Recognition Motif (RRM) of this splicing factor bound to an SMN2 derived RNA and showed, in combination with in vivo splicing assays, that contrary to what was previously proposed, this protein binds SMN2 at a specific site in exon 7. This discovery was important because it allowed us to propose a model of how hnRNP G and its interacting partner Tra2-b1 are co-recruited on SMN2 exon 7. Moreover, it allows the investigation of novel structure-based therapeutic strategies targeting those two strong splicing activators of SMN2 exon 7. In contrast to humans, only ~4% of Saccharomyces cerevisiae genes contain introns, and very few among them undergo alternative splicing. S. cerevisiae lacks classical SR proteins but contains 3 SR-like proteins. Npl3 is the only SR-like protein in budding yeast that was reported to influence splicing. We identified the minimal RNA motifs recognized by the 2 RRMs of Npl3 and determined the structure of their individual complexes using NMR. The structures explained the specific interaction of Npl3 with RNA and allowed us to design point mutations affecting the binding of either RRM1 or RRM2 to RNA. Using those Npl3 variants, we showed that the interaction of both domains with RNA is required for the splicing function of Npl3 in yeast. However, the binding of RRM1 but not RRM2 to RNA is required for the interaction with chromatin remodeling complexes. Finally, we used a combined CRAC and iCLIP approach to show that the motifs bound by each RRM in vitro were enriched in the Npl3 binding sites in vivo. Based on all those results, we proposed a model explaining the binding of both RRMs to the natural RNA targets of Npl3. In summary, we investigated both structurally and functionally how two important splicing factors belonging to the two main families of splicing regulators bind specifically their RNA target sequences and how this binding contributes to their functions in vivo.